IWHR_AI_Lable_Floater_V1.zipIWHR_AI_Lable_Floater_V1： 用于检测内陆水域漂浮物的注释数据集和基准

海洋垃圾对海洋生态系统构成严重威胁，及时清除内陆水域的漂浮垃圾可有效防止漂浮垃圾入海。准确的物体检测系统是有效清除漂浮物的前提。然而，水中复杂的光线条件、小尺寸物体和其他因素对漂浮物体的检测构成了巨大的挑战。为了促进漂浮物污染问题的解决，促进人工智能技术在水务行业的应用，我们提出了第一个基于岸基拍摄设备从真实水域场景采集的水域漂浮数据集 IWHR_AI_Lable_Floater_V1。该数据集由 3000 张图像组成，其中包含准确的注释信息，以支持基于视觉的水面漂浮物检测任务。我们进行了大量基线实验，以评估主流对象检测算法在该数据集上的性能。结果表明，包括最先进的模型 YOLOv9 在内的模型的检测精度都很低，这也表明漂浮物体检测是一项具有挑战性的任务。

Marine debris poses a severe threat to marine ecosystems. Timely removal of floating waste from inland waters can effectively prevent such debris from entering the ocean. Accurate object detection systems are a prerequisite for effective floating waste removal. However, complex underwater lighting conditions, small-sized objects, and other factors pose significant challenges to floating object detection. To address the floating waste pollution problem and promote the application of artificial intelligence technologies in the water industry, we propose the first aquatic floating waste dataset IWHR_AI_Lable_Floater_V1, which is collected from real-water scenes using shore-based shooting equipment. This dataset consists of 3000 images with accurate annotation information, supporting vision-based surface floating waste detection tasks. We conducted extensive baseline experiments to evaluate the performance of mainstream object detection algorithms on this dataset. The results show that the detection accuracy of models including the state-of-the-art YOLOv9 is relatively low, which further demonstrates that floating object detection is a highly challenging task.